[PATCH] rt2500usb: don't mark register accesses as inline

[PATCH] rt2500usb: don't mark register accesses as inline

[Arnd Bergmann]

When CONFIG_KASAN is set, we get a rather large stack here:

drivers/net/wireless/ralink/rt2x00/rt2500usb.c: In function 'rt2500usb_set_device_state':
drivers/net/wireless/ralink/rt2x00/rt2500usb.c:1074:1: error: the frame size of 3032 bytes is larger than 100 bytes [-Werror=frame-larger-than=]

If we don't force those functions to be inline, the compiler can figure this
out better itself and not inline the functions when doing so would be harmful,
reducing the stack size to a merge 256 bytes.

Note that there is another problem that manifests in this driver, as a result
of the typecheck() macro causing even larger stack frames.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 drivers/net/wireless/ralink/rt2x00/rt2500usb.c | 19 +++++--------------
 1 file changed, 5 insertions(+), 14 deletions(-)

diff --git a/drivers/net/wireless/ralink/rt2x00/rt2500usb.c b/drivers/net/wireless/ralink/rt2x00/rt2500usb.c
index 62357465fe29..0d2670a56c4c 100644
--- a/drivers/net/wireless/ralink/rt2x00/rt2500usb.c
+++ b/drivers/net/wireless/ralink/rt2x00/rt2500usb.c
@@ -55,7 +55,7 @@ MODULE_PARM_DESC(nohwcrypt, "Disable hardware encryption.");
  * If the csr_mutex is already held then the _lock variants must
  * be used instead.
  */
-static inline void rt2500usb_register_read(struct rt2x00_dev *rt2x00dev,
+static void rt2500usb_register_read(struct rt2x00_dev *rt2x00dev,
 					   const unsigned int offset,
 					   u16 *value)
 {
@@ -66,7 +66,7 @@ static inline void rt2500usb_register_read(struct rt2x00_dev *rt2x00dev,
 	*value = le16_to_cpu(reg);
 }
 
-static inline void rt2500usb_register_read_lock(struct rt2x00_dev *rt2x00dev,
+static void rt2500usb_register_read_lock(struct rt2x00_dev *rt2x00dev,
 						const unsigned int offset,
 						u16 *value)
 {
@@ -77,16 +77,7 @@ static inline void rt2500usb_register_read_lock(struct rt2x00_dev *rt2x00dev,
 	*value = le16_to_cpu(reg);
 }
 
-static inline void rt2500usb_register_multiread(struct rt2x00_dev *rt2x00dev,
-						const unsigned int offset,
-						void *value, const u16 length)
-{
-	rt2x00usb_vendor_request_buff(rt2x00dev, USB_MULTI_READ,
-				      USB_VENDOR_REQUEST_IN, offset,
-				      value, length);
-}
-
-static inline void rt2500usb_register_write(struct rt2x00_dev *rt2x00dev,
+static void rt2500usb_register_write(struct rt2x00_dev *rt2x00dev,
 					    const unsigned int offset,
 					    u16 value)
 {
@@ -96,7 +87,7 @@ static inline void rt2500usb_register_write(struct rt2x00_dev *rt2x00dev,
 				      &reg, sizeof(reg));
 }
 
-static inline void rt2500usb_register_write_lock(struct rt2x00_dev *rt2x00dev,
+static void rt2500usb_register_write_lock(struct rt2x00_dev *rt2x00dev,
 						 const unsigned int offset,
 						 u16 value)
 {
@@ -106,7 +97,7 @@ static inline void rt2500usb_register_write_lock(struct rt2x00_dev *rt2x00dev,
 				       &reg, sizeof(reg), REGISTER_TIMEOUT);
 }
 
-static inline void rt2500usb_register_multiwrite(struct rt2x00_dev *rt2x00dev,
+static void rt2500usb_register_multiwrite(struct rt2x00_dev *rt2x00dev,
 						 const unsigned int offset,
 						 void *value, const u16 length)
 {
-- 
2.9.0

Re: [PATCH] rt2500usb: don't mark register accesses as inline

[Kalle Valo]

Arnd Bergmann <arnd@arndb.de> writes:

> When CONFIG_KASAN is set, we get a rather large stack here:
>
> drivers/net/wireless/ralink/rt2x00/rt2500usb.c: In function 'rt2500usb_set_device_state':
> drivers/net/wireless/ralink/rt2x00/rt2500usb.c:1074:1: error: the frame size of 3032 bytes is larger than 100 bytes [-Werror=frame-larger-than=]
>
> If we don't force those functions to be inline, the compiler can figure this
> out better itself and not inline the functions when doing so would be harmful,
> reducing the stack size to a merge 256 bytes.
>
> Note that there is another problem that manifests in this driver, as a result
> of the typecheck() macro causing even larger stack frames.
>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

[...]

> -static inline void rt2500usb_register_multiread(struct rt2x00_dev *rt2x00dev,
> -						const unsigned int offset,
> -						void *value, const u16 length)
> -{
> -	rt2x00usb_vendor_request_buff(rt2x00dev, USB_MULTI_READ,
> -				      USB_VENDOR_REQUEST_IN, offset,
> -				      value, length);
> -}
> -

I guess you removed this because it was not actually used anywhere? Just
checking.

-- 
Kalle Valo

Re: [PATCH] rt2500usb: don't mark register accesses as inline

[Arnd Bergmann]

On Wed, Feb 15, 2017 at 6:49 AM, Kalle Valo <kvalo@codeaurora.org> wrote:
> Arnd Bergmann <arnd@arndb.de> writes:
>
>> When CONFIG_KASAN is set, we get a rather large stack here:
>>
>> drivers/net/wireless/ralink/rt2x00/rt2500usb.c: In function 'rt2500usb_set_device_state':
>> drivers/net/wireless/ralink/rt2x00/rt2500usb.c:1074:1: error: the frame size of 3032 bytes is larger than 100 bytes [-Werror=frame-larger-than=]
>>
>> If we don't force those functions to be inline, the compiler can figure this
>> out better itself and not inline the functions when doing so would be harmful,
>> reducing the stack size to a merge 256 bytes.
>>
>> Note that there is another problem that manifests in this driver, as a result
>> of the typecheck() macro causing even larger stack frames.
>>
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>
> [...]
>
>> -static inline void rt2500usb_register_multiread(struct rt2x00_dev *rt2x00dev,
>> -                                             const unsigned int offset,
>> -                                             void *value, const u16 length)
>> -{
>> -     rt2x00usb_vendor_request_buff(rt2x00dev, USB_MULTI_READ,
>> -                                   USB_VENDOR_REQUEST_IN, offset,
>> -                                   value, length);
>> -}
>> -
>
> I guess you removed this because it was not actually used anywhere? Just
> checking.

Correct, if I just remove the 'inline' annotation, we get a warning about an
unused function here. Sorry for not mentioning that in the changelog.

    Arnd

Re: rt2500usb: don't mark register accesses as inline

[Kalle Valo]

Arnd Bergmann <arnd@arndb.de> wrote:
> When CONFIG_KASAN is set, we get a rather large stack here:
> 
> drivers/net/wireless/ralink/rt2x00/rt2500usb.c: In function 'rt2500usb_set_device_state':
> drivers/net/wireless/ralink/rt2x00/rt2500usb.c:1074:1: error: the frame size of 3032 bytes is larger than 100 bytes [-Werror=frame-larger-than=]
> 
> If we don't force those functions to be inline, the compiler can figure this
> out better itself and not inline the functions when doing so would be harmful,
> reducing the stack size to a merge 256 bytes.
> 
> Note that there is another problem that manifests in this driver, as a result
> of the typecheck() macro causing even larger stack frames.
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

Patch applied to wireless-drivers-next.git, thanks.

727241660912 rt2500usb: don't mark register accesses as inline

-- 
https://patchwork.kernel.org/patch/9572947/

Documentation about submitting wireless patches and checking status
from patchwork:

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches